[C++ code gen, dictionary] Make instance names more consistent

[bocchino]

When an instance I is defined in a module M, the name of the instance is M_I. However, the code generation is not entirely consistent:

1. In the dictionary, the name becomes I instead of M_I. This makes the name shorter, but it can create ambiguity because it deletes the module prefix. E.g., M_I and N_I have the same name I in the dictionary. We should be consistent and use M_I in this case; a user who wants the unqualified name I can define the instance I outside of any module scope.
2. In the generated C++, the name becomes I inside the namespace M and M_I outside the namespace M. This makes it hard to write initialization code that works in all cases. We should either (1) use M_I consistently in all cases or (2) use a C++ namespace qualifier M::I instead of M_I.

We can fix (1) as we develop the new JSON dictionary. We can fix (2) by updating fpp-to-cpp.

[jwest115]

For part 2 of this issue - there are a few ways to go about this:

• Use C++ qualification, i.e. M::M::i
• Use underscore qualification (what we currently have), i.e. M::M_i
  • Easier to see that topology qualifier is M and component instance name is M_i
• Make component instance names fully qualified - would be consistent with dictionary

Example:

module M {
  passive component C { }
  instance i: C base id 0x100
  topology T { 
     instance i 
  }
}

C++ qualification

namespace M {
  //! M::i
  extern C M::i;
}
M::M::i

vs.

Underscore qualification

namespace M {
  //! M_i
  extern C M_i;
}
M::M_i